Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Braided rivers from space: monitoring surface-water dynamics using spectral water indices

Kirin, Mara LU (2026) NGEK11 20261
Department of Earth and Environmental Sciences (MGeo)
Abstract
Braided rivers are among the most dynamic and ecologically valuable fluvial systems, yet their complexity makes remote sensing-based monitoring particularly challenging. The Tagliamento River in northeastern Italy faces potential future regulation through proposed flood-gate infrastructure near the Dignano Bridge, making the establishment of baseline conditions a priority. This study pursued two objectives: evaluating the suitability of index–threshold combinations for surface-water extraction and characterizing spatiotemporal variability in surface-water extent and persistence within a 6 km reach divided into upstream and downstream sections, during 2024–2025. Specifically, four spectral indices: the Normalized Difference Water Index... (More)
Braided rivers are among the most dynamic and ecologically valuable fluvial systems, yet their complexity makes remote sensing-based monitoring particularly challenging. The Tagliamento River in northeastern Italy faces potential future regulation through proposed flood-gate infrastructure near the Dignano Bridge, making the establishment of baseline conditions a priority. This study pursued two objectives: evaluating the suitability of index–threshold combinations for surface-water extraction and characterizing spatiotemporal variability in surface-water extent and persistence within a 6 km reach divided into upstream and downstream sections, during 2024–2025. Specifically, four spectral indices: the Normalized Difference Water Index (NDWI), Modified NDWI (MNDWI), Automated Water Extraction Index (AWEI), and Sentinel Multi-band Water Index (SMBWI) were tested in combination with manual, variance-based, and minimum cross entropy thresholding approaches. Performance varied considerably across index–threshold combinations, reflecting the high spectral complexity of the braided environment. Among the tested approaches, manually thresholded NDWI achieved highest classification accuracy (OA = 0.96, Kappa = 0.92), while SMBWI combined with minimum cross entropy thresholding produced the second-highest results (OA = 0.93, Kappa = 0.86) without requiring manual intervention. Consequently, manually thresholded NDWI was applied to characterize the spatiotemporal dynamics of the study reaches. Spatiotemporal analysis revealed pronounced temporal variability in wetted extent, with both reaches displaying a coefficient of variation of approximately 78%. Persistence mapping confirmed the dominance of intermittently inundated areas, with the largest proportion of wetted area occurring within the 4 to 8-month persistence class (~23%), while permanently wetted channels accounted for approximately 2% of each reach. Fine-scale hydrological connectivity could not be reliably assessed at 10 m resolution. The results demonstrate the potential of freely available multispectral imagery for long-term braided river monitoring while establishing a pre-regulation baseline for the Dignano reach. (Less)
Please use this url to cite or link to this publication:
author
Kirin, Mara LU
supervisor
organization
course
NGEK11 20261
year
type
M2 - Bachelor Degree
subject
keywords
Sentinel-2, surface-water extraction, spectral water indices, NDWI, braided river monitoring, channel dynamics, Tagliamento river
language
English
id
9238617
date added to LUP
2026-06-16 10:26:11
date last changed
2026-06-16 10:26:11
@misc{9238617,
  abstract     = {{Braided rivers are among the most dynamic and ecologically valuable fluvial systems, yet their complexity makes remote sensing-based monitoring particularly challenging. The Tagliamento River in northeastern Italy faces potential future regulation through proposed flood-gate infrastructure near the Dignano Bridge, making the establishment of baseline conditions a priority. This study pursued two objectives: evaluating the suitability of index–threshold combinations for surface-water extraction and characterizing spatiotemporal variability in surface-water extent and persistence within a 6 km reach divided into upstream and downstream sections, during 2024–2025. Specifically, four spectral indices: the Normalized Difference Water Index (NDWI), Modified NDWI (MNDWI), Automated Water Extraction Index (AWEI), and Sentinel Multi-band Water Index (SMBWI) were tested in combination with manual, variance-based, and minimum cross entropy thresholding approaches. Performance varied considerably across index–threshold combinations, reflecting the high spectral complexity of the braided environment. Among the tested approaches, manually thresholded NDWI achieved highest classification accuracy (OA = 0.96, Kappa = 0.92), while SMBWI combined with minimum cross entropy thresholding produced the second-highest results (OA = 0.93, Kappa = 0.86) without requiring manual intervention. Consequently, manually thresholded NDWI was applied to characterize the spatiotemporal dynamics of the study reaches. Spatiotemporal analysis revealed pronounced temporal variability in wetted extent, with both reaches displaying a coefficient of variation of approximately 78%. Persistence mapping confirmed the dominance of intermittently inundated areas, with the largest proportion of wetted area occurring within the 4 to 8-month persistence class (~23%), while permanently wetted channels accounted for approximately 2% of each reach. Fine-scale hydrological connectivity could not be reliably assessed at 10 m resolution. The results demonstrate the potential of freely available multispectral imagery for long-term braided river monitoring while establishing a pre-regulation baseline for the Dignano reach.}},
  author       = {{Kirin, Mara}},
  language     = {{eng}},
  note         = {{Student Paper}},
  title        = {{Braided rivers from space: monitoring surface-water dynamics using spectral water indices}},
  year         = {{2026}},
}